List your company

Visual Question Answering on InfoSeek (test)

57.9AccuracyQwen2-VL-Oracle
6.014419.484732.95546.4253May 29, 2023Nov 12, 2023Apr 28, 2024Oct 13, 2024Mar 29, 2025Sep 13, 2025Feb 28, 2026

Evaluation results

MethodLinks
Qwen2-VL-OracleLLM=Qwen2-7B, KB=Wiki, Setting=Zero-shot2025.0657.957.957.9—
MAD-RAGBackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0154.06———
DoLABackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0152.25———
AdaCADBackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0152.04———
VCDBackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0151.68———
MAD-RAGBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0151.59———
Vanilla RAGBackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0151.44———
ALFARBackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0151.44———
MAD-RAGBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0150.19———
MAD-RAGBackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0148.43———
ALFARBackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0148.09———
CADBackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0147.98———
DoLABackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0147.28———
VCDBackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0146.98———
CoRe-MMRAGLLM=Qwen2-7B, KB=Wiki, Setting=Fine-tuned2025.0646.545.246.9—
AdaCADBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0146.45———
AdaCADBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0146.45———
Vanilla RAGBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0146.33———
AdaCADBackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0146.33———
Vanilla RAGBackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0145.71———
SPINBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0145.35———
ALFARBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0145.11———
DoLABackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0144.69———
CADBackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0144.12———
OPERABackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0144.03———
Vanilla RAGBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0144.01———
VCDBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0143.86———
VCDBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0143.61———
OPERABackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0143.03———
DoLABackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0143.01———
Qwen2-VL-1-StageLLM=Qwen2-7B, KB=Wiki, Setting=Fine-tuned2025.064342.343.4—
CoRe-MMRAGLLM=Qwen2-7B, KB=Wiki, Setting=Zero-shot2025.0642.942.343.3—
SPINBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0140.97———
Qwen2-VL-1-StageLLM=Qwen2-7B, KB=Wiki, Setting=Zero-shot2025.0640.940.840.9—
VHRBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0140.19———
LLaVA-mR2AGLLM=Vicuna-7B, KB=Wiki, Setting=Fine-tuned2025.0639.439.139.7—
CADBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0139.06———
CADBackbone=LLaVA-1.5-13B, Context Setting=Oracle Contexts2026.0139.06———
ALFARBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0138.7———
VHRBackbone=LLaVA-1.5-7B, Context Setting=Oracle Contexts2026.0138.06———
Qwen2-VL-2-StageLLM=Qwen2-7B, KB=Wiki, Setting=Fine-tuned2025.0637.336.837.6—
Qwen2-VL-MMSTaRLLM=Qwen2-7B, KB=Wiki, Setting=Fine-tuned2025.0635.134.835.2—
Qwen2-VL-2-StageLLM=Qwen2-7B, KB=Wiki, Setting=Zero-shot2025.063534.935.1—
Qwen2-VL-MMSTaRLLM=Qwen2-7B, KB=Wiki, Setting=Zero-shot2025.0633.933.434—
RA-VQAv2 w/ PreFLMRRAG System=RA-VQAv2, Retriever=PreFLMR2026.0232.1———
V-Retrver-7BModel Scale=7B2026.0231.9———
Wiki-LLaVALLM=Vicuna-7B, KB=Wiki, Setting=Fine-tuned2025.0628.930.127.8—
LamRA-7BModel Scale=7B2026.0228.8———
VHRBackbone=Qwen2.5-VL-3B, Context Setting=Oracle Contexts2026.0128.16———
EchoSightLLM=LLAMA3-8B, KB=Wiki, Setting=Fine-tuned2025.0627.7———
RORA-VLMLLM=Vicuna-7B, KB=Wiki+Web, Setting=Fine-tuned2025.0626.927.325.1—
VHRBackbone=Qwen2.5-VL-7B, Context Setting=Oracle Contexts2026.0124.04———
MMA-RAGModel=Qwen2VL-7B2026.0223.9———
RIRModel=Qwen2VL-7B2026.0223.3———
P(true)Model=Qwen2VL-7B2026.0222.6———
Qwen2-VL-ParamLLM=Qwen2-7B, Setting=Zero-shot2025.0622.123.121.8—
PaLI-XOCR pipeline input=false, Resolution=224x2242023.0521.8———
CLIPModel=Qwen2VL-7B2026.0220.9———
CoTModel=Qwen2VL-7B2026.0220.4———
MMA-RAGModel=Idefics2-8B2026.0220.3———
Zero shotModel=Qwen2VL-7B2026.0219.9———
Closed-book (parametric)Backbone=Qwen2.5-VL-7B, Context Setting=Parametric only2026.0119.58———
MMA-RAGModel=Idefics3-8B2026.0218.1———
Closed-book (parametric)Backbone=Qwen2.5-VL-3B, Context Setting=Parametric only2026.0117.88———
Method [48]OCR pipeline input=false, Resolution=224x2242023.0517.7———
RIRModel=Idefics3-8B2026.0217.5———
RIRModel=Idefics2-8B2026.0217.2———
CLIPModel=Idefics2-8B2026.0216.4———
Few shotModel=Qwen2VL-7B2026.0215.9———
CoTModel=Idefics2-8B2026.0215.5———
Zero shotModel=Idefics2-8B2026.0215.1———
CoTModel=Idefics3-8B2026.0214.3———
Few shotModel=Idefics2-8B2026.0214.2———
P(true)Model=Idefics2-8B2026.0214.1———
P(true)Model=Idefics3-8B2026.0212.9———
CLIPModel=Idefics3-8B2026.0212.6———
Zero shotModel=Idefics3-8B2026.0212.5———
LLaVA-1.5LLM=Vicuna-7B, Setting=Zero-shot2025.0612.21310.3—
Few shotModel=Idefics3-8B2026.028.5———
Closed-book (parametric)Backbone=LLaVA-1.5-13B, Context Setting=Parametric only2026.018.14———
Closed-book (parametric)Backbone=LLaVA-1.5-7B, Context Setting=Parametric only2026.018.01———
EchoSightGenerator Model=Mistral-7B | LLaMA3-8B, Gen. FT=false, Ret. FT=true2026.04———31.3
LLM-RAGenerator Model=BLIP2-Flan-T5XL, Gen. FT=true, Ret. FT=true2026.04—26.1220.923.14
mR2AGGenerator Model=LLaVA-1.5-7B, Gen. FT=true, Ret. FT=true2026.04—40.639.840.2
OMGMGenerator Model=LLaVA-1.5-7B, Gen. FT=true, Ret. FT=true2026.04—43.4643.5343.49
ReflectiVAGenerator Model=LLaVA-MORE-8B, Gen. FT=true, Ret. FT=true2026.04—40.439.840.1
RoRA-VLMGenerator Model=LLaVA-1.5-7B, Gen. FT=true, Ret. FT=false2026.04—27.3425.126.9
Wiki-LLaVAGenerator Model=LLaVA-1.5-7B, Gen. FT=true, Ret. FT=false2026.04—30.127.828.9
WikiSeekerGenerator Model=Qwen3-VL-8B-Instruct, Gen. FT=false, Ret. FT=true2026.04—38.737.6938.19
WikiSeekerGenerator Model=Qwen2.5-7B-Instruct, Gen. FT=false, Ret. FT=true2026.04—35.9934.9335.46
WikiSeekerGenerator Model=Inspector+Qwen2.5-7B-Instruct, Gen. FT=true, Ret. FT=true2026.04—43.8245.6444.72